A novel method for an unbiased estimate of cross-ancestry genetic correlation using individual-level data

Abstract Cross-ancestry genetic correlation is an important parameter to understand the genetic relationship between two ancestry groups for a complex trait. However, existing methods cannot properly account for ancestry-specific genetic architecture, which is diverse across ancestries, producing biased estimates of cross-ancestry genetic correlation. Here, we present a method to construct a genomic relationship matrix (GRM) that can correctly account for the relationship between ancestry-specific allele frequencies and ancestry-specific causal effects. Through comprehensive simulations, we show that the proposed method outperforms existing methods in the estimations of SNP-based heritability and cross-ancestry genetic correlation. The proposed method is further applied to six anthropometric traits from the UK Biobank data across 5 ancestry groups. One of our findings is that for obesity, the estimated genetic correlation between African and European ancestry cohorts is significantly different from unity, suggesting that obesity is genetically heterogenous between these two ancestry groups.